Active Learning using compact exported model from Customvision.ai

[Andrew Jansen]

Describe the solution you'd like
I would like the active learning custom url and custom file path options to work with a model trained in Microsofts CustomVision.ai i.e. the url to function when pointing to a general model iteration published in the CustomVision.ai platform and a compact model exported as Tensorflow.js to function when pointed at using the custom file path option.

Additional context
This will enable greater sharing and collaboration of models and workload to multiple people, research institutions and agencies, to generate labelled datasets.